Predicting the emergence of antibiotic resistance through multi-omics approaches and immune system surveillance

Awardee Organization: BOSTON COLLEGE
Contact PI / Project Leader: VAN OPIJNEN, TIM
Project Number: 1U01AI124302-01


The rise of antibiotic resistance in bacterial populations reflects inadequate counterselection by the antibiotic itself, by host immunity, or by fitness costs of the resistance mechanism. Our inability to control resistance stems from limited understanding of these three forces of selection and especially of the interplay between antibiotic dosage, how bacteria populations respond to antibiotics, and host immunity. This project maps these interactions in high definition for the bacterial pathogens Streptococcus pneumoniae and Acinetobacter baumannii, both of which are serious threats and can cause antibiotic-resistant pneumonia. Specifically, in this project: 1) RNA-seq is applied to construct transcriptional networks; 2) Two complementary, genome-wide methods of screening mutations are used to map the phenotypic networks that affect resistance. One screening method called Tn-seq screens all possible gene knockout mutations for their roles in antibiotic susceptibility and resistance. The other method, experimental evolution, enables mutants to arise naturally and compete for success in the antibiotic condition. These interactions will be studied in vitro, using standard culture and under resistance-inducing conditions, including biofilms on plastic surfaces that are often the source of nosocomial A. baumannii infections. Mutant responses to antibiotic selection will also be studied in vivo, in which bacterial populations infect mice that re treated with antibiotic but vary in their immune competence; 3) The state-of-the-immune-system (sIS) will be profiled during mouse infections to identify sIS fingerprints reflecting both immune success and failure, when the bacterial population evades both antibiotic and immune pressure. All of these methods will be applied to define the bacterial-adaptive and host-response pathways for 20 different clinically relevant antibiotics; 4) Ribo-seq is used to evaluate evolved strains and the manner in which their interactions with the immune system changes over time. Finally, all of these bacterial- networks, immune-states and responses will be integrated into a joint model that will be learnt using a novel plug-and-play toolbox for fast prototyping of data driven solutions. This model will be validated by testing the likelihood of the emergence of resistance through: 1) selection in the presence of five novel antibiotics, 2) selection in the presence of two different resistant pathogens, and 3) through selection in the presence of three different immunocompromised hosts. Therefore, the ultimate goal of this project is to design a `plug-and-play learning toolbox’ that is able to forecast the likelihood of the emergence of antibiotic resistance and prioritize antibiotic and immune therapies that enable more rapid, effective treatment with minimal risk of treatment failure.